Regulation of Bcl-xl Channel Activity by Calcium*

Recent studies have demonstrated that the anti-apoptotic proteins, Bcl-2 and Bcl-xl, with the carboxyl-terminal hydrophobic domain removed, form cation-selective channels in the lipid bilayer reconstitution system. However, the regulatory properties of these channels are unknown. In this study, we investigated the ion-conducting properties of full-length Bcl-xl in the lipid bilayer reconstitution system. Our findings indicate that Bcl-xl forms a cation-selective channel that conducts sodium but not calcium and that Bcl-xl channel activity is reversibly inhibited by luminal calcium with a half-dissociation constant of ∼60 μm. This calcium-dependent regulation of the Bcl-xl channel provides new insights into the roles of calcium and Bcl-2-related proteins in the programmed cell death pathway.

Recent studies have demonstrated that the anti-apoptotic proteins, Bcl-2 and Bcl-xl, with the carboxyl-terminal hydrophobic domain removed, form cation-selective channels in the lipid bilayer reconstitution system. However, the regulatory properties of these channels are unknown. In this study, we investigated the ionconducting properties of full-length Bcl-xl in the lipid bilayer reconstitution system. Our findings indicate that Bcl-xl forms a cation-selective channel that conducts sodium but not calcium and that Bcl-xl channel activity is reversibly inhibited by luminal calcium with a half-dissociation constant of ϳ60 M. This calcium-dependent regulation of the Bcl-xl channel provides new insights into the roles of calcium and Bcl-2-related proteins in the programmed cell death pathway.
Bcl-2 and its related proteins are critical regulators of programmed cell death or apoptosis (1,2). Composed of both antiand pro-apoptotic members, this family of proteins functions in a programmed cell death pathway common to most multicellular organisms. It is generally believed that the ratio of death antagonists, such as Bcl-2 and Bcl-xl, and death agonists, such as Bax and Bcl-xs, plays a major role in the fate of the cell following an apoptotic stimulus (3). The anti-apoptotic proteins have also been suggested to function by interacting with caspases/CED-4 homologs (4,5). Most of the Bcl-2-related proteins contain a hydrophobic carboxyl-terminal sequence that anchors the protein to membranes of organelles, including the endoplasmic reticulum (ER), 1 mitochondria and nucleus, which play important roles in apoptosis (6 -9). Despite their importance in regulating development and homeostasis in multicellular organisms, the exact physiological function(s) of these proteins remain elusive.
One of the current theories is that Bcl-2 family members regulate ion fluxes. This idea is supported by evidence that Bcl-2 overexpression prevents calcium redistribution from the ER to mitochondria following growth factor withdrawal (10) and inhibits apoptosis-associated calcium waves (11) and nuclear calcium uptake (12). Bcl-2 overexpression has also been shown to enhance the uptake of calcium by mitochondria (13) and preserve mitochondrial transmembrane potential (14). Our previous data have demonstrated that Bcl-2 overexpression is associated with a reduction in the transient elevation of cytosolic calcium induced by thapsigargin-mediated ER calcium ATPase inhibition (15). Furthermore, Bcl-2 overexpression maintains calcium homeostasis and calcium-dependent protein processing in the ER of thapsigargin-treated cells (16).
Recently, the x-ray and NMR structure of Bcl-xl was shown to resemble the physical structure of ion channel-forming bacterial toxins, such as diphtheria toxin and colicin Ia (17). Subsequently, Bcl-xl, Bcl-2 and Bax, have been shown to form ion channels using the lipid bilayer reconstitution system (18 -21). However, these studies all utilized proteins from which the carboxyl-terminal hydrophobic domain had been deleted. Previous studies of the channel forming bacterial toxin, colicin Ia, showed that the channel formed by protein which lacks the hydrophobic transmembrane domain differs in conductance from the one formed by the intact colicin Ia (22). Thus, it is presently unclear whether the ion channel forming properties of the truncated Bcl-2 family members necessarily correspond to those of the intact proteins. Furthermore, it seems paradoxical that Bcl-2 and Bcl-xl, which function to inhibit apoptosis, should structurally and functionally resemble bacterial toxins whose main purpose is to form pores in cell membranes and ultimately destroy cells by disrupting ion homeostasis. This suggests that the ability of Bcl-2 and Bcl-xl to promote cell survival may not derive solely from their ability to form a channel. An explanation to this paradox may lie in understanding how Bcl-2 and Bcl-xl channel activity is regulated.
In light of our recent evidence that depletion of cellular calcium abrogates the anti-apoptotic effect of Bcl-2 (16), we set out to examine whether calcium is able to regulate the channel activity of anti-apoptotic Bcl-2 family members in a lipid bilayer reconstitution system. In the present report, we show that full-length Bcl-xl, produced in a bacterial expression system, forms a monovalent cation-selective channel that conducts sodium but not calcium. Significantly, we observed that Bcl-xl channel activity is inhibited by calcium, indicating for the first time that Bcl-xl forms a calcium-regulated cation channel.

MATERIALS AND METHODS
Plasmid Preparation-A full-length human bcl-xl cDNA (23) was polymerase chain reaction-amplified, using flanking primers (forward ϭ 5Ј-CCCCCTTCTAGAATGTCTCAGAGCAACCGG-3Ј and reverse ϭ 5Ј-GGGGCATGCCTCGAGTCATTTCCGACTGAAGAGTGAG-CC-3Ј, start/stop codons are underlined) to introduce a XbaI site in the 5Ј end and a XhoI site in the 3Ј end. The XbaI and XhoI-digested, gel-purified polymerase chain reaction product was subcloned into the pProex-1 expression vector (Life Technologies, Inc.) containing a 6-histidine (6ϫHis) sequence, which allows affinity purification of expressed 6ϫHis-Bcl-xl fusion protein. The entire bcl-xl cDNA sequence was identical to the published sequence with the exception of two conservative changes, A70G and V152A.
Protein Purification-A single colony of Escherichia coli, transformed with the full-length bcl-xl cDNA, was cultured at 37°C in LB medium with 100 g/ml ampicillin. Induction was carried out at an A 600 of 0.7 with 1 mM isopropyl-␤-D-thiogalactopyranoside at 37°C for 5 h before harvesting cells by centrifugation. The cells were pelleted and resuspended in 2-5 volumes of sonication buffer (50 mM Na 3 PO 4 (pH 8.0), 300 mM NaCl, and 1 mM phenylmethylsulfonyl fluoride). Following sonication (1-min bursts/1-min cooling/200 -300 watts), the cells were centrifuged at 20,000 ϫ g for 1 h at 4°C. The pellet was resuspended and solubilized as described previously by Loo and Clarke (24) with minor modifications. Briefly, the pellet was resuspended in 0.3 ml of buffer A (50 mM Na 3 PO 4 (pH 8.0), 500 mM NaCl, 50 mM imidazole, and 20% (v/v) glycerol). The proteins were then solubilized at 4°C by adding 1 ml of solubilization buffer (buffer A ϩ 1% (w/v) CHAPS). Insoluble material was removed by centrifuging at 16,000 ϫ g for 15 min. The supernatant was applied to the nickel-nitrilotriacetic acid column that was pre-equilibrated with 0.5 ml of nickel-agarose (Qiagen) in buffer B (buffer A ϩ 0.1% (w/v) CHAPS). The column was washed twice with 0.6 ml of buffer B and with 0.6 ml of buffer C (10 mM Tris-Cl (pH 7.0), 500 mM NaCl, 80 mM imidazole (pH 7.0), 0.1% (w/v) CHAPS, and 20% (v/v) glycerol). The 6ϫHis-tagged Bcl-xl proteins were eluted with 0.25 ml of buffer D (10 mM Tris-Cl (pH 7.0), 500 mM NaCl, 300 mM imidazole (pH 7.0), 0.1% (w/v) CHAPS, and 20% (v/v) glycerol) in multiple fractions. Typically, cells from 100 ml of culture yielded 0.5-3.0 g of Bcl-xl protein. Purified proteins were then characterized by SDS-PAGE (12.5% gels), followed by silver staining (Bio-Rad) and Western blotting.
Western Blot Analysis-Immunoblots were performed as described previously (16). Briefly, full-length 6ϫHis-tagged Bcl-xl proteins were mixed with the sample buffer (200 mM Tris-Cl (pH 6.7), 10% SDS, 5% ␤-mercaptoethanol, 15% glycerol, 0.01% bromphenol blue) and separated on a 12.5% linear SDS-PAGE. The proteins were then transferred to a nitrocellulose membrane and blotted with the Bcl-xl polyclonal antibody (Santa Cruz Biotechnology) and horseradish peroxidaselinked secondary antibody using the ECL detection system (Amersham Pharmacia Biotech).
Planar Lipid Bilayer Preparation and Single Channel Recordings-Planar lipid bilayer membranes were formed across an aperture of 200 m diameter with a mixture of phosphatidylethanolamine/phosphatidylserine/cholesterol in a ratio of 5:5:1 as described previously (25). Single channel currents were recorded with an Axopatch 200A patch clamp unit, and data analyses were performed with pClamp Software. The eluted fraction containing the purified Bcl-xl protein was either used directly or concentrated by Centriplus-30 (Amicon) and reconstituted in liposomes (phosphatidylethanolamine/phosphatidylserine/cholesterol in a ratio of 1:1:1) for the single channel measurements. The two preparations gave similar results. Purified Bcl-xl (60 -100 ng) or Bcl-xl containing liposome was added to the cis-solution in the presence of asymmetric NaCl conditions at pH 7.4. The experiments were performed at room temperatures (23-25°C). Single channel currents were recorded at 1 ms/point with the cut-off filter frequency set at 1 KHz.

RESULTS
The full-length Bcl-xl protein was expressed in E. coli following isopropyl-␤-D-thiogalactopyranoside induction and purified to homogeneity using nickel-nitrilotriacetic acid affinity chromatography. The Western blot shown in Fig. 1A represents the elution profile from the nickel column. The purified Bcl-xl has an apparent molecular mass of ϳ31 kDa, which corresponds to the endogenous Bcl-xl (data not shown). It has been shown previously that Bcl-xl runs slightly higher than the predicted molecular mass of 26 kDa for the full-length Bcl-xl protein (18). Silver staining of an eluted fraction revealed a single band, indicating that the preparation was Ͼ95% pure (Fig. 1B.) To characterize the single channel function of full-length Bcl-xl, purified protein was reconstituted in the lipid bilayer, and the selected current traces shown in Fig. 2 were recorded with 200 mM NaCl present in the cis-solution (to which the Bcl-xl protein was added) and 50 mM NaCl present in the trans-solution. Bcl-xl displays channel activity, producing a larger current at ϩ40 mV than that at Ϫ40 mV ( Fig. 2A). In addition, the current-voltage relationship has a reversal potential of Ϫ30 mV (Fig. 2B), which corresponds to a selectivity ratio of P Na /P Cl ϭ 13.9. These findings indicate the cationselective nature of the Bcl-xl channel. In separate experiments with a recording solution consisting of 200 mM NaCl (cis)/200 mM KCl (trans), the full-length Bcl-xl channel had a reversal potential of ϳϪ5 to 0 mV, which corresponds to a relative selectivity ratio of P Na /P K ϭ ϳ1-1.2. This value is similar to the truncated Bcl-xl channel reported by Minn et al. (18). A characteristic feature of the Bcl-xl channel is the appearance of multiple conductance states, as shown in Fig. 2C. Typically, six distinctive conductance levels could be identified, with each conductance level of ϳ50 picosiemens. The association of multiple conductance states with the full-length Bcl-xl channel is similar to that of the Bcl-xl channel that had the carboxylterminal hydrophobic domain deleted (18). This characteristic may be because of the propensity of Bcl-xl proteins to interact with one another, as this prevents us from distinguishing the current measurements from either single channels or multiple channels. At the present, we do not know the stoichiometry of the Bcl-xl channel, i.e. the number of subunits that participates in the formation of the cation-selective channel.
To confirm that the conductance shown in Fig. 2 was a property of the Bcl-xl protein, we measured its conductance in the presence of the Bcl-xl polyclonal antibody, which recognizes first 19 amino acids of the amino terminus. As shown in Fig. 3, addition of ϳ31 nM antibody to the cis-solution produced a significant reduction in Bcl-xl channel activity (n ϭ Ͼ3), whereas addition of an equivalent amount of bovine serum albumin to the cis-solution was without effect (data not shown). Furthermore, when the antibody was added to the trans-solution, no change in channel activity was observed. These findings suggest that the Bcl-xl channel activity may be regulated by protein-protein interactions at the amino terminus and confirm that the channel activity observed in the planar lipid bilayer system was indeed produced by Bcl-xl rather than any minor contaminants in the protein preparation. In addition, these data confirm that the protein is oriented in the ciscytoplasmic trans-luminal manner.
To study whether the full-length Bcl-xl channel is regulated by calcium, 100 mM CaCl 2 was added to the trans-solution. The current traces shown in Fig. 4 were obtained with a NaCl gradient of 200 (cis)/50 mM (trans). Following the addition of 100 mM CaCl 2 to the trans-solution, the activities of the channel at both ϩ60 mV and Ϫ60 mV were reduced to essentially zero, indicating that calcium inhibited the function of the Bcl-xl channel. Furthermore, because an inward current was not observed at Ϫ60 mV, the Bcl-xl channel does not appear to be permeable to calcium.
The dose-response relationship between luminal calcium concentration and Bcl-xl channel activity was examined, to further understand the effect of luminal calcium. The results indicate that calcium attenuates both the current amplitude and opening probability of the Bcl-xl channel activity, with a K i   FIG. 1. Purification of full-length Bcl-xl protein. The full-length human bcl-xl cDNA was cloned into the pProex-1 vector, with a 6ϫHis added to the 5Ј end. The Bcl-xl protein was purified using a nickelaffinity column. The proteins were analyzed by SDS-PAGE on a 12.5% gel, followed by Western blot. A: lane 1, solubilized extracts after high speed centrifugation; lane 2, fraction eluted with 50 mM imidazole; lane 3, fraction eluted with 80 mM imidazole; lanes 4 -8, fractions eluted with 300 mM imidazole. The fraction containing the highest amount of Bcl-xl protein was silver-stained, which revealed a single band of ϳ31 kDa, indicating that the protein was Ͼ95% pure (B). of 57.78 Ϯ 26.14 M (Fig. 5A). In addition, the inhibitory effect was because of the action of luminal calcium, as suggested by the following observations. First, adding calcium to the cissolution had a lesser inhibitory effect on the Bcl-xl channel current amplitude and opening probability (Fig. 5A). Second, adding 10 mM MgCl 2 to the trans-solution did not affect the channel activity (Fig. 5A). In addition, the inhibitory effect of the Bcl-xl channel activity by luminal calcium was readily and consistently reversed by adding EGTA to the trans-solution (n ϭ 5), as illustrated in Fig. 5B. DISCUSSION Our findings indicate that full-length Bcl-xl forms a monovalent cation selective channel, with conductance properties similar to those described previously using recombinant proteins from which the carboxyl-terminal hydrophobic domain   had been deleted (18 -21). Despite the evidence suggesting that Bcl-2-related proteins conduct ions, the relationship of this property to apoptosis induction or repression is presently unknown. In the present study, however, we show that Bcl-xl channel activity can be regulated by calcium, a finding that may provide new insights into the function of the Bcl-2 family in the apoptotic pathway.
Regulation of ion channels by calcium is a well documented phenomenon. For example, intracellular calcium regulates the rate of calcium-dependent inactivation of the voltage-gated L-type calcium channel as well as the open-closed transitions of a wide variety of potassium channels, such as Maxi-K (Ca) (26,27). Also, changes in luminal calcium concentration are known to regulate the rate of calcium release from the sarcoplasmic reticulum vesicles of muscle cells, as well as the single channel opening probability of the ryanodine receptor/calcium release channels (28).
Thus, our observation that an anti-apoptotic protein such as Bcl-xl is regulated by luminal calcium is not unusual, yet novel as it may have a significant impact on how we view the function of these proteins in the apoptotic pathway. Although calcium has been implicated as a mediator of apoptosis, its exact role is enigmatic. However, the maintenance of a high level of calcium within the ER lumen is known to be essential for cellular functions, including translation (29), protein processing (30), cell division (31), and cell survival. Recently, Tsien and colleagues (32) demonstrated, with their "cameleons" fluorescent indicators, that the calcium concentration in the ER of individual cells is as high as 400 M under physiological conditions. In addition, using ER-targeted, low calcium affinity aequorin reconstituted with coelenterazine n, the physiological concentration of calcium in the ER lumen is estimated to be between 500 and 600 M in intact cells (33). Our findings indicate that luminal calcium in the high M range inhibits Bcl-xl channel activity. This suggests that when the calcium concentration in the ER lumen is maintained at this level, the Bcl-xl channel remains in a closed state, supporting ion homeostasis.
Unlike the ER lumen, the free calcium concentration is relatively low in mitochondria. For example, mitochondrial free calcium in a single living cardiac myocyte is estimated to be in the high nanomolar range (34). Our data suggest that calcium at this concentration would keep the Bcl-xl channel open. However, what do Bcl-xl and its channel activity in the mitochondria have to do with promoting cell survival? Interestingly, a recent study by Shimizu and colleagues demonstrated that Bcl-2 prevents loss of mitochondrial membrane potential (⌬) induced by 20 M calcium in isolated mitochondria (35). Their findings suggested that Bcl-2 maintained this ⌬ by mediating proton efflux. However, at a higher calcium concentration (100 -200 M), Bcl-2 failed to mediate proton efflux, leading to the collapse of the ⌬. This is consistent with our findings regarding the dose-response relationship between luminal calcium concentration and Bcl-xl channel activity. Hence, it is likely that Bcl-xl channel activity may be regulated by the luminal calcium concentration of the organelle with which it is associated, thereby contributing to its anti-apoptotic function.
Based on the present findings, whether Bax and other proapoptotic proteins can be regulated by calcium merits immediate investigation.